Zou Hanting, Lan Tianmeng, Jiang Yongwen, Yu Xiao-Lan, Yuan Haibo
Tea Research Institute, The Chinese Academy of Agricultural Sciences, Hangzhou 310008, China.
Foods. 2024 Nov 21;13(23):3718. doi: 10.3390/foods13233718.
As a crucial stage in the processing of black tea, rolling plays a significant role in both the color transformation and the quality development of the tea. In this process, the production of theaflavins, thearubigins, and theabrownins is a primary factor contributing to the alteration in color of rolled leaves. Herein, tea pigments are selected as the key quality indicators during rolling of black tea, aiming to establish rapid detection methods for them. A machine vision system is employed to extract nine color feature variables from the images of samples subjected to varying rolling times. Then, the tea pigment content in the corresponding samples is determined using a UV-visible spectrophotometer. In the meantime, the correlation between color variables and tea pigments is discussed. Additionally, Z-score and PCA are used to eliminate the magnitude difference and redundant information in original data. Finally, the quantitative prediction models of tea pigments based on the images' color features are established by using PLSR, SVR, and ELM. The data show that the Z-score-PCA-ELM model has the best prediction effect for tea pigments. The Rp values for the model prediction sets are all over 0.96, and the RPD values are all greater than 3.50. In this study, rapid determination methods for tea pigments during rolling of black tea are established. These methods offer significant technical support for the digital production of black tea.
作为红茶加工的关键阶段，揉捻对茶叶的色泽转化和品质形成都具有重要作用。在此过程中，茶黄素、茶红素和茶褐素的生成是导致揉捻叶色泽变化的主要因素。本文选取茶叶色素作为红茶揉捻过程中的关键品质指标，旨在建立其快速检测方法。采用机器视觉系统从不同揉捻时间的样品图像中提取9个颜色特征变量。然后，使用紫外可见分光光度计测定相应样品中的茶叶色素含量。同时，探讨颜色变量与茶叶色素之间的相关性。此外，利用Z分数和主成分分析（PCA）消除原始数据中的量纲差异和冗余信息。最后，采用偏最小二乘回归（PLSR）、支持向量回归（SVR）和极限学习机（ELM）建立基于图像颜色特征的茶叶色素定量预测模型。数据表明，Z分数-PCA-ELM模型对茶叶色素的预测效果最佳。模型预测集的Rp值均大于0.96，RPD值均大于3.50。本研究建立了红茶揉捻过程中茶叶色素的快速测定方法。这些方法为红茶的数字化生产提供了重要的技术支持。
